lichess.org

Solving the game of chess

As technology improves, do you think about engine will be created that plays the absolute best moves? So it is impossible to beat this engine? Only draw at best?

We now have 7 men tablebases. When we have 32 men tablebases the game is solved.

Every ten years we have a "men" more. So we are lucky if see something like the 10,11,12-men TB in our lifetime. Storage place for 7-men is hardly to handle on your private PC, I guess it's about factor 10.000 higher per men.

At some stage, the curve may flatten out, may it not? 32 piece egtb may not be much more difficult than 31 piece egtb.

I do doubt that a man more trend will continue. Problem grows exponentially. 8 piece would take about 140 PB of storage (googles whole storage is about 1000 PB) and 9 piece would take 140 000 PB., or 140 Exa-B And computational resources would alsi grow at same rate. So we might see 8 eventually but 9 piece is probably beyond what is doable. Or at least is too expensive to do. Perhaps some trully partial tablebases might show up

Chess could be solved within a decade if someone devoted the time and resources.

I know some people will say there are 80 gazillion possible games or whatever but how many of those are simple mate in 1's where you make 50,000 random king moves. The tablebases could be drastically condensed by just teaching engines to recognize a completely won game.

@Savage47

-chess IS already solved „weakly“. Humans don’t stand a chance
-it won’t be solved „strongly“ (=completely) in next few centuries

Every 10 years a man more: 32 - 7 = 25, 25 x 10 = 250. So expect it solved in 250 years. There may be breakthroughs in technology: both storage technology and processor speed. I remember in the 80's people spoke about a futuristic 3M computer: it would have a million pixels, 1 megabyte harddisk and 1 million operations per second. Now we easily are at 3G. So every 10 years capacity and computing power goes x10, just by evolving technology. Totally new technology like quantum computing or clustering may make it even faster. I expect chess solved by 2100.

computer tech as we know it is reaching it limits. transistor cannot be smaller thatn they are as size of molecules and un-predictability due quantum effects just gets too high
www.eetasia.com/news/article/SPIE-Conference-Predicts-Bumpy-Chip-Roadmap
same goes information storage capacity.

One sign of this that CPU speed has not increase during last 10 years. Additional speed has gained by add ing parallel capcity. which obviously has limits as well

So mankind does have limits imposed by nature.

Strong solution to chess is probably beyond possibilities as -math term- it implies all responses to all moves have been verified. No matter how silly. And number of possible chess games exceeds number of atoms in visible universe. So even if make assumption that we store one bit/atom we coudl not do it.

Practically solved sure. Within 10 years or so all top computer games will end in draw and further progress cannot be made

@petri999

I remember hearing once that a Commissioner of the patent office in the 1800s said the patent office should be closed because everything that could be invented had already been invented.

20-25 years a desktop computer with a 1gb hard drive would have been impressive now I've got a phone I carry around with me that has 8gb of RAM and a half a terabyte hard drive.

Things that seem impossible now may not always be that way. Quantum computers, for one, could change everything. And as I said earlier, you could still prove a solution at a tiny fraction of space by eliminating meaningless moves. If the game is 1. g4, e5 2. f3 what is gained by analyzing 80 billion moves after 2...a6 and then 2...a5 and so on? 99.99% of the potential moves in chess are meaningless to analyze and a solution could be proven without analyzing them.

Reconnecting